Phys Chem Minerals (1999) 26: 576±583
Ó Springer-Verlag 1999
ORIGINAL PAPER
E. Huang á J.-F. Lin á J. Xu á T. Huang
Y.-C. Jean á H.-S. Sheu
Compression studies of gibbsite and its high-pressure polymorph
Received: 28 September 1998 / Revised, accepted: 22 December 1998
Abstract Various X-ray di€raction methods have been
applied to study the compression behavior of gibbsite,
Al(OH)3, in diamond cells at room temperature. A phase
transformation was found to take place above 3 GPa
where gibbsite started to convert to its high-pressure
polymorph. The high-pressure (HP) phase is quenchable
and coexists with gibbsite at the ambient conditions after
being unloaded. This HP phase was identi®ed as nordstrandite based on the di€raction patterns obtained at
room pressure by angle dispersive and energy dispersive
methods. On the basis of this structural interpretation,
the bulk modulus of the two polymorphs, i.e., gibbsite
and nordstrandite, could be determined as 85 ‹ 5 and
70 ‹ 5 GPa, respectively, by ®tting a Birch-Murnaghan
equation to the compression data, assuming their K0o as
4. Molar volume cross-over occurs at 2 GPa, above
which the molar volume of nordstrandite is smaller than
that of gibbsite. The di€erences in the molar volume and
structure between the two polymorphs are not signi®cant, which accounts for the irreversibility of the phase
transition. In gibbsite, the axial compressibility behaves
as c/co > a/ao > b/bo. This is due to the fact that the
dioctahedral sheets along the c-axis are held by the relatively weak hydrogen bonding, which results in the
greater compressibility along this direction. In nordstrandite, the axial compressibility is b/bo > c/co > a/
ao, which can also be interpreted as resulting from the
the existence of hydrogen bonds along the b-axis.
E. Huang (&) á J. Xu á T. Huang
Institute of Earth Sciences, Academia Sinica,
P.O. Box 1-55, Nankang, Taipei, Taiwan, R.O.C.
J.-F. Lin
Geophysical Science, University of Chicago,
Chicago, IL., U.S.A.
Y.-C. Jean á H.-S. Sheu
Synchrotron Radiation Research Center,
Hsinchu, Taiwan, R.O.C.
Introduction
The water reservoir in the Earth's mantle may be dominately hosted in the hydrous minerals (e.g., Akimoto
and Akaogi 1984). Therefore, the study of hydrous
minerals at high-pressure and temperature is crucial for
understanding the dynamic processes of water circulation in the mantle. Examples such as triggering deep
focus earthquakes (Kirby 1987; Meade and Jeanloz
1991) and regulating the water budget have been documented in hydrous minerals under mantle environments.
In the past few years, some hydrous minerals with
trioctahedral layer structure such as brucite (Mg(OH)2)
(e.g., Du€y et al. 1995; Xia et al. 1998) and portlandite
(Ca(OH)2) (Desgranges et al. 1993; Pavese et al. 1997)
have been studied by the high-pressure experiments from
which important geophysical implications are documented. These hydrous minerals show di€erent behaviors at high pressures, such as amorphization in
Ca(OH)2 (Meade and Jeanloz 1990) and dehydration in
Mg(OH)2 (Johnson and Walker 1993). In our previous
works, we have devoted ourselves to the studies of the
high-pressure behaviors of hydrous and anhydrous aluminum oxides (Huang et al. 1995; Xu et al. 1995; Huang
et al. 1996). Some of these compounds, such as diaspore,
AlO(OH), do not show a signi®cant change in structure
but yield unexpected variations in Raman modes at high
pressures (Huang et al. 1995). On the other hand,
gibbsite, Al(OH)3, shows drastic changes in both Raman
and di€raction patterns at high pressures (Huang et al.
1996). The results obtained from the Raman spectroscopic studies are of interest when compared with those
of other OH-bearing minerals at high pressure (Kruger
et al. 1989). Therefore, it is obvious that more highpressure research on compounds in the Al2O3-H2O
system should be conducted in order to understand the
behavior of the OH-bearing phases and the phase relationships of minerals in the system.
Gibbsite with a space group of P21/n has an analogous structure as brucite but possesses dioctahedral
577
layers in contrast to those trioctahedral layers in magnesian hydrous phases. Huang et al. (1996) reported that
gibbsite undergoes a polymorphic phase transition
above 3 GPa based on Raman and X-ray observations.
The high-pressure phase was tentatively determined to
be nordstrandite (space group P1) based on the diffraction patterns of the sample yielded from the unloading process. Unfortunately, their high-pressure
X-ray data were inadequate to justify whether this highpressure phase is indeed nordstrandite. Therefore, we
have carried out further in situ high-pressure study on
gibbsite and also a more detailed structural identi®cation of the high-pressure phase. Furthermore, the compressibility can be determined for the two phases.
Energy dispersive X-ray di€raction (EDXRD) has
routinely been used as a tool in the study of materials
under pressures with synchrotron X-ray radiation. Recently, much progress has been made in the detecting
system of the angle dispersive X-ray di€raction (ADXRD) method. Among all the detectors used in the
ADXRD method, the image plate (IP) has proven to be
the most ecient way for the collection of the di€raction
signals. IP is a photosimulatable ¯uorescent plate, which
is a digitally readable X-ray photograph. It has the
characteristics of high absorption coecient, sensitivity
and greater dynamic range than the conventional ADXRD detectors and it also has the capability of being
easily read out (Shimomura et al. 1992). Hence, IP has
proven to be a very ecient detector for the angle dispersive X-ray di€raction in diamond anvil cell (DAC)
experiment (Huang 1996). Moreover, the data can be
digitized for structural analysis, which makes this
method suitable for high-pressure di€raction research
especially when synchrotron radiation is used (Chen and
Takemura 1993). The application of IP to the X-ray
studies and structural re®nement of samples in diamond
anvil cells were successfully carried out by Shimomura
et al. (1992) and Nelmes et al. (1992). In this report, we
will demonstrate the use of EDXRD and ADXRD
methods in studying the pressure-induced phase transition in Al(OH)3 at room temperature.
Experimental methods
The Baker Company's product of synthetic gibbsite powder, which
was the same sample used by Huang et al. (1996), was used as the
starting material for all experiments in this study. Both EDXRD
and ADXRD methods were used to study the compression of
gibbsite at room temperature. The experimental set up is described
below.
ADXRD method
Two di€erent ADXRD methods were used in this study. Their
main goal is to examine the gibbsite sample that was quenched to
the ambient conditions after it was compressed beyond the transition pressure. The two ADXRD methods di€er in the X-ray source
and the detector used for the collection of the di€raction signal.
In the ®rst method, gibbsite powder was con®ned by a stainless
steel gasket and then compressed in a diamond cell up to a nominal
pressure of 5 GPa. The sample was then unloaded and mounted on
a sample holder in a Debye-Scherrer camera (with a diameter of
57.3 mm) in which a photographic ®lm was inserted. The X-ray
source used was CoKa radiation with running conditions of 45 KV
and 30 mA. It took 24 hours for the exposure of the ®lm to collect
the di€raction signal.
In the second method, the image plate (IP) was used in coupled
with the synchrotron radiation of the Synchrotron Radiation Research Center (SRRC), Hsin-Chu, Taiwan. A monochromatic
beam of 14.5 KeV was used as the X-ray source. The sample was
loaded in the hole (250 lm diameter) of a gasket made of stainless
steel and then slightly compressed between the diamond anvil cell
so as to ®ll up the sample chamber. The cell was then mounted and
aligned to bring the collimated X-ray beam (normally of the order
of 100 lm) to fall on the sample in the diamond cell. The sample
was then compressed to approximately 6 GPa and quenched to
room pressure. The di€raction patterns of the sample at highpressure and quenched condition were collected by the image plate
with an exposure time of 24 hours. Optical alignment using a
monitor that traced the image of the sample ensured that the position of the sample did not change during the process of removing
and remounting of the sample.
EDXRD method
Two sets of runs were conducted in the EDXRD experiments. One
was to identify the quenched phase and to correlate with the results
obtained by the ADXRD method. The other was to determine the
compressibility of the two polymorphs of Al(OH)3. The white radiation component of synchrotron radiation was used as the X-ray
source for both runs.
In the quenching experiment, the gibbsite sample was loaded in
the gasket hole of a stainless steel and compressed in a diamond cell
above the phase transition pressure. Several di€raction patterns
were taken to monitor the process of the transition. The pressure
was then decreased to the ambient conditions and the di€raction
pattern of the quenched phase was taken. No pressure calibrant
was loaded in the cell because the run was designed to take the
di€raction pattern of the quenched phase without interfering peaks
from pressure calibrant. The run was carried out at the B2 station
of Cornell High Energy Synchrotron Source, Cornell University.
Except for the 2h angle (taken at 14°), the experimental geometry
and data acquisition time are similar to those of the compressibility
run described below.
In the compressibility run, the sample was mechanically mixed
with gold powder and loaded in a diamond anvil cell. A T301
stainless steel was used as the gasket with a diameter of 200 lm for
the sample chamber. Methanol-ethanol (4:1) solution was used as
pressure medium to maintain a hydrostatic condition. The experiments were conducted at beamline X-17C, National Synchrontron
Light Source, Brookhaven National Laboratory. An energy dispersive X-ray di€raction pattern was taken each time after the
pressure increment was made. An intrinsic Ge detector was set at
an angle (2h) of 7° and the acquisition time for each spectrum was
about 10 minutes. The X-ray beam size was limited to 30 ´ 30 lm.
Gold was used as pressure calibrant (Heinz and Jeanloz 1984) and
the average error in pressure measurements was ‹0.3 GPa. All of
the spectra were treated by Jandel Scienti®c Peak-Fit Program for
the position of di€raction peaks.
Results
Quenched phase
Figure 1 shows the di€raction patterns of the quenched
product obtained by the methods using ®lm, image plate
and Ge-detector as recording systems. The di€raction
pattern obtained by the photographic ®lm is shown in
578
material is shown in Fig. 1e. The d-spacings corresponding to all of the 4 di€raction patterns in Fig. 1 are
listed in Table 1 along with the peaks of gibbsite and
nordstrandite according to the JCPDS cards. By comparing these di€raction peaks with those of gibbsite and
nordstrandite, it is concluded that the di€raction patterns of the quenched sample can be interpreted as a
mixture of both gibbsite and nordstrandite (Table 1).
Compressibility measurement
Figure 2 shows a series of the EDXRD spectra of
Al(OH)3 taken during the loading and unloading processes. At pressure lower than about 4 GPa, the X-ray
patterns showed ten di€raction peaks of gibbsite, i.e.,
the (002), (110), (200), (20
2), (11
2), (21
1), (
112), (311),
(021) and (024). The d-spacings of these peaks decrease
with pressure up to about 4 GPa. At 5.2 GPa, some new
peaks started to show up and co-existed with those of
gibbsite. The di€raction peaks of gibbsite disappeared
Fig. 1a±e The di€raction patterns of quenched sample obtained by
a ®lm b image plate and d EDXRD methods. c is the conversion from
the di€raction rings to intensity versus d-spacing plot of b. e is the
di€raction pattern of the starting material obtained by the EDXRD
method. The di€raction peaks resolved in these patterns are shown as
numbers and their corresponding d-spacings are listed in Table 1
Fig. 1a. The original two dimensional print-out of the
di€raction patterns obtained by the image plate is shown
in Fig. 1b, which was processed and shown as an ordinary di€ractometric chart in Fig. 1c. The EDXRD
pattern of the quenched Al(OH)3 is shown in Fig. 1d.
For comparison, the EDXRD pattern of the starting
Fig. 2 A series of EDXRD patterns showing the phase transition in
Al(OH)3 with the increase (loading) and decrease (unloading) in
pressure. Di€raction peaks of gibbsite and the high-pressure phase are
denoted as g and n, respectively
579
Table 1 Comparison of the di€raction patterns of the starting material and the quenched sample of Al(OH)3 with standard gibbsite and
nordstrandite from the JCPDS cardsa
Starting material
Quenched sample
EDXRD
EDXRD
IP
D-S
(hkl)
d (AÊ)
I/Io
(hkl)
d (AÊ)
I/Io
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
G/N
?
G/N
G/N
N
N
G
G
G
?
G/N
G/N
G/N
G
G
G
G/N
G/N
G/N
G
G
G
G
G/N
N
G/N
G
N
G/N
G/N
?
4.811 [26]
4.728 [60]
4.356 [100]
±
3.978 [50]
±
3.324 [4]
3.200 [8]
3.128 [3]
2.718 [3]
2.466 [30]
2.406 [12]
±
2.239 [27]
2.189 [8]
2.068 [4]
1.990 [12]
±
±
1.746 [6]
1.706 [2]
±
1.623 [±]
1.594 [3]
1.513 [2]
1.477 [8]
1.464 [14]
±
1.432 [2]
1.405 [2]
1.306 [3]
4.823 [s]
4.701 [m]
4.345 [s]
4.242 [s]
4.011 [m]
3.885 [w]
3.330 [m]
3.210 [w]
3.098 [w]
±
2.464 [s]
2.389 [m]
±
2.228 [s]
2.191 [m]
2.050 [s]
1.965 [m]
1.876 [w]
1.780 [w]
1.723 [w]
1.662 [w]
1.647 [w]
1.628 [w]
1.602 [w]
1.503 [w]
1.479 [m]
1.472 [m]
1.440 [w]
1.420 [w]
1.408 [w]
1.312 [w]
(002)
±
(110)
(200)
±
(202†
(112†
(211,112)
(202,103)
±
(311)
(213, 311)
(312)
(022, 213)
(312)
(204, 313)
(023)
(411)
(314)
(024)
(323)
(420)
(421)
(315, 422)
±
(±)
(±)
±
(±)
(±)
±
4.82
±
4.34
4.30
±
3.35
3.31
3.17
3.08
±
2.44
2.37
2.28
2.23
2.15
2.03
1.98
1.90
1.79
1.74
1.67
1.65
1.63
1.58
±
1.477
1.457
±
1.441
1.409
±
[100]
±
[40]
[20]
±
±
[6]
[8]
[4]
±
[16]
[20]
[4]
[6]
[8]
[12]
[10]
[8]
[10]
[10]
[10]
[4]
[2]
[4]
±
[2]
[3]
±
[2]
[2]
±
(002)
±
(110†
(110)
(200)
(202†
(112†
(112)
(112)
(312)
(310)
(004, 310)
(022)
±
±
±
(314)
(312)
(024)
±
±
±
±
(316)
(314)
(602)
±
(330)
(224)
(330)
±
4.79
±
4.33
4.22
4.16
3.896
3.604
3.446
3.022
2.481
2.454
2.393
2.265
±
±
±
2.015
1.904
1.781
±
±
±
±
1.595
1.513
1.478
±
1.440
1.431
1.403
±
[100]
±
[25]
[25]
[20]
[15]
[10]
[10]
[15]
[15]
[10]
[35]
[35]
±
±
±
[30]
[20]
[15]
±
±
±
±
[10]
[10]
[10]
±
[20]
[5]
[10]
±
4.839 [7]
4.726 [5]
4.356 [38]
±
±
±
3.328 [27]
3.184 [22]
3.109 [100]
±
2.456 [65]
2.383 [72]
2.288 [11]
2.242 [10]
2.162 [21]
2.051 [27]
1.994 [19]
1.917 [11]
1.805 [15]
1.751 [14]
1.681 [9]
1.647 [4]
±
1.586 [5]
±
±
1.457 [9]
±
1.440 [10]
1.410 [7]
±
4.834 [6]
4.705 [10]
4.347 [30]
±
3.959 [17]
±
3.319 [17]
3.188 [12]
3.120 [7]
2.696 [3]
2.450 [100]
2.385 [40]
2.278 [19]
2.221 [47]
2.160 [8]
2.051 [20]
1.975 [31]
1.910 [6]
1.803 [9]
1.749 [11]
1.680 [5]
±
1.634 [2]
1.581 [5]
1.499 [2]
±
1.448 [15]
±
±
1.407 [4]
±
Gibbsite
Nordstrandite
a
The unit for all d-spacings is in AÊ. G and N stand for gibbsite and
nordstrandite, respectively. The numbers, xx, in [xx] indicate the
relative intensity of the peak compared with the most intense peak
in the pattern. In the ®lm method, the intensity of the di€raction
peaks is represented by letters: s, m and w, stand for strong,
medium and weak, respectively
when the pressure was brought up to 5.5 GPa. Little
change was observed when the sample was further
compressed to a maximum pressure of 9.6 GPa. During
the unloading process, the high-pressure phase persisted
even at the ambient conditions.
During the loading process, the d-spacing of each
di€raction peak can be calculated from the Ed value
which was determined to be 71.14 KeV á AÊ at a ®xed 2h
of 14°. The pressure of each run was determined from
the equation of state of cold (Heinz and Jeanloz 1984)
with its molar volume calculated based on the d-spacings
of Au (111), (200), (220) and (311) peaks. The variation
of d-spacing of each di€raction peaks versus pressure is
shown in Fig. 3. The scheme for indexing the di€raction
peaks above 4 GPa is discussed in the next section.
of d-spacing of each peak with pressure (Fig. 3) that a
phase transformation in gibbsite has occurred above
4 GPa. In our previous EDXRD experiments carried
out at the Photon Factory, Tsukuba, Japan, we also
observed a phase transition took place (Huang et al.
1996). However, the exact pressure at which the phase
transition occurred was not well de®ned because the
pressure increment was too large (Huang et al. 1996).
Subsequent Raman spectroscopic observation has revealed that gibbsite transforms to a high-pressure phase
at 3 GPa (between 2.2 and 3.2 GPa; Huang et al. 1996).
In the present EDXRD study, the phase transition was
found to take place between 3.4 GPa and 5.2 GPa. This
transition pressure is close to but still higher than that
observed by the Raman spectroscopy. Since we have no
data points between 3.4 and 5.2 GPa, the determination
of the transition pressure relies on a comparison in relative intensity of the di€raction peaks between gibbsite
and its HP phase. Judging from the relative intensity
between n(002) and g(002) (Fig. 2), which represent the
high-pressure and gibbsite phase, respectively, we think
that the transition pressure should be much lower than
5.2 GPa where the HP phase has already reached its
Discussion
Phase transformation in Al(OH)3
It is clearly seen from the change in the di€raction
patterns (Fig. 2) and the drastic change in the variation
580
Fig. 3 The variations of d-spacings of all di€raction peaks with
pressure. The phase transition is observed to take place between 3.4
and 5.2 GPa. The di€raction peaks are indexed and shown in the
Figure. The open circles are the data taken during the unloading
process
mature stage. However, even a lower value of 4 GPa is
taken as the transition pressure, it still makes the value
higher than that determined based on the Raman spectroscopic observation. Since the two polymorphs have
very similar di€raction patterns but drastically di€erent
Raman spectra, it is likely that the structural change
during the transition is not easily detected as the change
in the vibrational modes (Huang et al. 1996). Therefore,
the pressure at which transition occurs based on X-ray
di€raction lags behind that of the Raman observation.
High-pressure phase of Al(OH)3
The high-pressure phase of gibbsite has never been determined structurally in the past. From Figs. 1 and 2, it
is clearly seen that the quenched phase is di€erent from
the starting material. Therefore, it is reasonable to check
the di€raction patterns of the quenched phase with those
of the polymorphs of Al(OH)3. Although gibbsite has
two other polymorphs, doyleite and bayerite, in addition
to nordstrandite (Chao et al. 1985), we think the quenched HP phase of gibbsite ®ts best to nordstrandite.
Among the 31 di€raction peaks detected in the quenched
sample (Table 1), some of them are the di€raction peaks
from the starting phase, gibbsite, and some of them can
be interpreted as the mixture of both gibbsite and nordstrandite. The positive evidence for the presence of
nordstrandite is provided by the di€raction peaks
numbered as 5, 6, 25 and 28, which correspond to the
(200), (20
2), (314) and (3
30) of nordstrandite. The
agreement in the value of d-spacing is in general within
‹0.005 AÊ, which is within the resolving power for image
plate and EDXRD method. It is also in reasonable
agreement with the ®lm result despite the fact that some
of the peaks show deviation in d-spacing that exceeds
the uncertainty of the ®lm method. However, since most
of the di€raction peaks can be indexed as the mixture of
gibbsite and nordstrandite, treating nordstrandite as the
candidate for the HP phase of gibbsite is by no means
fortuitous. It is noted that the relative intensity of the
di€raction peaks of the quenched high-pressure phase
are not fully in agreement with those of nordstrandite.
The intensities of di€raction peaks of the material are
a€ected by several factors such as the preferred orientation of the powdered sample, the interference of the
coexisting phases, the intensity pro®le of the incident
X-ray beam and the relaxation of the sample after being
unloaded from the diamond cell. Therefore, the match in
peak position should be a better criterion for correlating
the di€raction peaks than relative intensity.
Further evidence that the HP phase is nordstrandite
is found in high-pressure in situ compression experiments. Figure 4 shows the di€raction patterns of
gibbsite obtained by the IP and EDXRD methods under
a nominal pressure of 5 to 6 GPa. The di€raction pattern of gibbsite under a pressure of 5.5 GPa (using gold
as pressure calibrant) is also shown in Fig. 4d for comparison. The patterns show clearly the (002), (110),
(200)/(20
2), (11
2)/(
112), (
312), (02
2) and (31
4) di€raction peaks of nordstrandite. Although the relative intensity among these peaks is di€erent, the peak positions
match well with each other. Therefore, we conclude that
the high-pressure phase of gibbsite is nordstrandite.
Compressibility measurements
The variations in the d-spacing of each di€raction line of
gibbsite and its HP phase, nordstrandite, are shown in
Fig. 3. Once the structure of both phases in Al(OH)3 are
identi®ed, the variations of the cell parameters and
molar volume with pressure of these phases can be determined. We have used the di€raction lines of (002),
(110), (200), (20
2), (11
2), (21
1)/(
112), (31
1)/(021) and
(024) for the determination of lattice parameters of
gibbsite up to 5 GPa. For nordstrandite, the di€raction
peaks of (002), (1
10), (200)/(20
2), (11
2)/(
112), (312),
(02
2) and (
314) were used for the calculation of the
lattice parameters of nordstrandite above 5 GPa. We
have adopted the method of Novak and Colville (1989)
for the calculation of the lattice parameters. The com-
581
Fig. 5 The variations of a/ao, b/bo, c/co and V/Vo of two polymorphs
of Al(OH)3 with pressure. The molar volume compression curves
represent a bulk modulus of 85 and 70 GPa for gibbsite and
nordstrandite, respectively
Fig. 4a±d The di€raction patterns of gibbsite obtained by a image
plate at a nominal pressure of 6 GPa; c EDXRD method at a nominal
pressure of 5 GPa; and d EDXRD method at 5.5 GPa. b is the
conversion from the di€raction rings to intensity versus d-spacing plot
of a. Note that c and d were taken at di€erent 2h angles
pression data of these two phases are listed in Table 2
and plotted in Fig. 5. The bulk modulus of gibbsite and
nordstrandite can be determined by ®tting a BirchMurnaghan equation of state to the data as 85 ‹ 5 GPa
and 70 ‹ 5 GPa, respectively, assuming a ®rst pressure
derivative, K0o , as 4 in both cases.
In gibbsite, the axial compressibility behaves as c/
co > a/ao > b/bo below 4 GPa. This is due to the
dioctahedral sheets along the c axis, which are held by
relatively weak hydrogen bonding, resulting in the
largest compressibility along this direction (Fig. 6a).
Along the a- and b-axis, the octahedral Al-(OH)6 units
link together by sharing edges and thus form stronger
bonds than that along the c-axis. In nordstrandite, the
axial compressibility is b/bo >c/co > a/ao, which can
also be interpreted as due to the contribution of the
hydrogen bonds along the b-axis while stronger bonds
exist along the a- and c-axis where polyhedra are shared
(Fig. 6b). The anisotropic compressibility is commonly
observed in layered hydrous minerals in which bonds
with non-uniform strength exist. Similar anisotropic
compressibility is also found in diaspore (Xu et al.
1994). The present results also indicate that the compressibility of the high-pressure phase, nordstrandite, is
slightly greater than that of its low-pressure phase,
gibbsite.
The cell volume of the two phases of Al(OH)3 can be
calculated from their cell parameters since their structures are indenti®ed. It is noted that the molar volume of
nordstrandite (32.90 cm3/mol) is slightly larger than that
of gibbsite (31.48 cm3/mol) at the ambient conditions
(calculated from Table 2, with Z=8). Therefore, it is
expected that nordstrandite should be less stable than
582
Table 2 The compression data
of two Al(OH)3 polymorphsa
Gibbsite
Nordstrandite
P(GPa)
a/a0
b/b0
c/co
b
V/Vo
P(GPa) a/a0
b/b0
c/co
V/Vo
0
0.97
1.39
1.90
2.30
2.84
3.38
5.23
1.000
0.997
0.997
0.994
0.990
0.988
0.984
0.981
1.000
1.000
0.997
0.996
0.994
0.993
0.989
0.990
1.000
0.993
0.994
0.995
0.989
0.988
0.987
0.976
94.8
95.2
95.2
95.2
95.2
95.8
96.0
96.4
1.000
0.989
0.988
0.984
0.973
0.968
0.959
0.945
5.53
7.25
8.05
8.55
8.99
9.22
9.60
0
0.974
0.967
0.966
0.963
0.961
0.960
0.958
1.000
0.983
0.978
0.976
0.974
0.973
0.972
0.970
1.000
0.935
0.917
0.911
0.903
0.901
0.899
0.891
1.000
0.985
0.980
0.977
0.975
0.974
0.974
0.972
1.000
The lattice parameters of gibbsite at room pressure are ao=8.607 AÊ, bo=5.055 AÊ, co=9.645 AÊ
and b=94.84°, the unit cell volume Vo=418.1 AÊ3; those of nordstrandite are ao=8.890 AÊ,
bo=4.978 AÊ, co=10.539 AÊ and b=94.84°, the unit cell volumeVo=437.2 AÊ3. Uncertainities in pressure reading, cell parameter ratio, beta angle and molar volume ratio are ‹0.25 GPa, ‹0.001, ‹0.2°
and ‹0.003, respectively
a
Fig. 6a, b The crystal structure
of a gibbsite and b nordstrandite showing the arrangement
of the polyhedral layers. The
most compressible direction is
along the c-axis and b-axis of
gibbsite nordstrandite where
the polyhedral layers are held
by relatively weak hydrogen
bonds (modi®ed from Chao
et al. 1985)
gibbsite at room pressure as pointed out by Schoen and
Roberson (1970). In fact, gibbsite is the predominant
Al(OH)3 phase found in nature while nordstrandite is
sporadically found in regions where the pH value is high
(Schoen and Roberson 1970). However, when the variations of molar volume with pressure are compared for
these two phases, it is interesting to see that nordstrandite has the same molar volume as gibbsite at about
2 GPa. As the pressure increases, the di€erence in molar
volume between the gibbsite and its high-pressure
polymorph becomes greater. Therefore, the phase transition from gibbsite to nordstrandite is thermodynamically favored above 2 GPa. Because the crystal
structures of gibbsite and nordstrandite di€er only in the
stacking sequences of the dioctahedral sheets (Chao
et al. 1985) and the molar volumes di€er only slightly, it
is likely that the high-pressure phase could be quenched
at room pressure due to the diculty in removing the
remaining nordstrandite, during the process of regaining
the atomic arrangement of the original phase. Epitaxial
growth might have occurred during the phase transition
when the two phases are structurally inter-related. The
stress inherited during the initial growth of the HP phase
in the matrix of low-pressure phase would cause the
anomalous distortion in the lattice parameters of the HP
phase (Bassett and Huang 1987). Therefore, it is likely
that when nordstrandite was ®rst produced from
gibbsite, the lattice parameters would show anomalous
distortion. However, the resolution of the present results
is not good enough for the observation of this phenomenon. A detailed high-pressure single-crystal
compression study on gibbsite would provide more information for the mechanism of the gibbsite-nordstrandite phase transition.
Acknowledgements The writers wish to express their thanks to
the sta€ in the SRRC, CHESS and BNL for their persistent
help during the experimental runs. We owe our thanks to
Mr. Y.C. Yang, for his assistance in the computer programming
work. This work was a follow-through of the discovery of
Mr. Chen and Miss Li who had provided basic observations in
high-pressure behavior of Al(OH)3. This work was supported by
the National Science Council, NSC 85-2111-M-001-021 and NSC
86-2613-M-213-015.
References
Akimoto S, Akaogi M (1984) Possible hydrous magnesian silicates
in the mantle transition zone. In: Sunagawa I (ed) Material
Science of the Earth's Interior. Terra Sci. Publ. Co., Tokyo,
pp 477±480
583
Bassett WA, Huang E (1987) Mechanism of the body-centered
cubic-hexagonal closed-packed phase transition in iron. Science
238: 780
Chao GY, Baker J, Sabina AP, Roberts AC (1985) Doyleite, a new
polymorph of Al(OH)3, and its relationship to bayerite, gibbsite
and nordstrandite. Can Mineral 23: 21±85
Chen JH, Takemura K (1993) Two-dimensional angle-dispersive
X-ray di€raction by an image plate with MAX80, 34th HighPressure Conference. Rev High Pressure Sci Tech Spec Issue 2:
34±35
Desgranges L, Grebille D, Calvarin G, Chevrier G, Floquet N,
Niepce JC (1993) Hydrogen thermal motion in calcium hydroxides: Ca(OH)2. Acta Crystallogr B49: 812±817
Du€y TS, Meade C, Fei Y, Mao HK, Hemley RJ (1995) Highpressure phase transition in brucite, Mg(OH)2. Am Mineral 80:
222±230
Heinz DL, Jeanloz R (1984) The equation of state of gold calibration standard. J Appl Phys 55: 885±893
Huang E (1996) Applications of Image Plate in high-pressure research (I): phase transitions in some dioxides. Report to National Science Council, R.O.C.
Huang E, Lin JF, Xu J, Yu SC (1995) Raman spectroscopic study
of diaspore up to 25 GPa. J Geol Soc China 38: 25±36
Huang E, Li A, Xu J, Chen R, Yamanaka T (1996) High-pressure
phase transition in Al(OH)3: Raman and X-ray observations.
Geophys Res Lett 23: 3083±3086
Johnson MC, Walker D (1993) Brucite [Mg(OH)2] dehydration
and the molar volume of H2O to 15 GPa. Am Mineral 78: 271±
284
Kirby SH (1987) Localized polymorphic phase transformations in
high-pressure faults and applications to the physical mechanism
of deep earthquakes. J Geophys Res 92: 13789±13800
Kruger MB, Williams Q, Jeanloz R (1989) Vibrational spectra of
Mg(OH)2 and Ca(OH)2 under pressure. J Chem Phys 91: 5910±
5915
Meade C, Jeanloz R (1990) Static compression of Ca(OH)2 at room
temperature: observations of amorphization and equation of
state measurements to 10.7 GPa. Geophys Res Lett 17: 1157±
1160
Meade C, Jeanloz R (1991) Deep-focus earthquakes and recycling
of water into the Earth's mantle. Science 252: 68±72
Nelmes RJ, Hatton PD, McMahon MI, Piltz RO, Crain J, Cernik
RJ, Bushnell-Wye G (1992) Angle-dispersive powder-di€raction techniques for crystal structure re®nement at high pressure.
Rev Sci Instrum 63: 1039±1042
Novak GA, Colville AA (1989) A practical interactive least-square
cell-parameter program using an electronic spread-sheet and a
personal computer. Am Mineral 74: 488±490
Pavese A, Catti M, Ferraris G, Hull S (1997) P-V equation of state
of portlandite, Ca(OH)2, from powder neutron di€raction data.
Phys Chem Minerals 24: 85±89
Schoen R, Roberson CE (1970) Structures of aluminum hydroxide
and geochemical implications. Am Mineral 55: 43±77
Shimomura O, Takemura K, Fujihisa H, Fujii Y, Ohishi Y, Kikegawa T, Amemiya Y, Matsushita T (1992) Application of an
image plate to high-pressure x-ray study with a diamond anvil
cell. Rev Sci Instrum 63: 967±973
Xia X, Weidner DJ, Zhao H (1998) Equation of state of brucite:
single-crystal Brillouin spectroscopy study and polycrystalline
pressure-volume-temperature measurement. Am Mineral 83:
68±74
Xu J, Hu J, Ming LC, Huang E, Xie H (1994) The compression of
diaspore, AlO(OH) at room pressure up to 27 GPa. Geophys
Res Lett 21: 161±164
Xu J, Huang E, Lin JF, Xu Y (1995) Raman study at high pressure
and the thermodynamic properties of corundum: application to
Kie€er's model. Am Mineral 80: 1159±1167